Search

We argue for an extension of the proposal that grammars are in part shaped by processing systems. Hawkins (2014) and others who have advanced this idea focus primarily on parsing. Our extension focuses on production, and we use that to explore explanations for certain subject/object asymmetries in extraction structures. The phenomenon we examine, which we term the mirror asymmetry , runs in opposite directions for within-clause and across-clause (long-distance) extraction, showing a preference for subject extraction in the former and for object extraction in the latter. We review several types of evidence suggesting that the mirror asymmetry and related phenomena are best explained by an account of the formation of grammars that assigns an important role to properties of sentence planning in production.

Psycholinguistic studies of language processing have revolved historically around “modular” and “interactive” accounts of language use. Experimental reports diverge in claims for the penetration of non-linguistic background information on processing for sentence comprehension. Syntactic processing effects can persist despite available contextual constraints that are sufficient to resolve temporary ambiguity or garden path errors. Nevertheless, there are multiple reports of interactive effects between basic sentence processing and both semantic and non-linguistic contextual information. The chapter suggests a rationalization of such conflicting findings in standard psycholinguistic and experimental pragmatic research, relying on interactions between language comprehension systems and language production systems. Production processes are designed to incorporate discourse and environmental constraints on linguistic formulation. These may be used to filter the products of comprehension mechanisms. A key feature of the argument for complementary roles of the two systems is a degree of modular processing for syntax to be found in both systems.

Since early experimental explorations of pragmatic phenomena it has been documented that novel and established utterances are processed differently. This is especially relevant to processing of a class of utterances called “implicitures” (Bach, 1994) in which some aspects of content are not explicitly expressed by the words used—they are implicit. It has been suggested that at least some implicitures have become “standardized” for their content (Bach, 1998; Garrett and Harnish, 2007). That is, the standard use of these expressions conveys the relevant content even though the words uttered do not present that content as conventional, linguistic meaning. While some studies suggest that the implicitures are mandatorily inferred regardless of context (Bach, 1998), others claim that impliciture processing is context-dependent (Sperber and Wilson, 1986). We investigated this issue using spatial, temporal and possession implicitures in two reaction time experiments. Implicitures were presented context-free or embedded in contexts that either supported their preferred interpretation or cancelled it. The results indicated that implicitures are readily available when no context is provided and are produced even when context forces an alternative interpretation. These findings support a standardization view for at least some impliciture processing. Possible differences in processing mechanisms across theories of impliciture processing and across impliciture types are discussed.

This paper argues for broader consideration of children’s language production systems and, in that context, describes research on children’s planning of syntactic structures. The research presented here measures non-fluency patterns in elicited utterances of varied syntactic type. We describe and interpret several regularities in these patterns for two groups of children (‘young’: three‐five-year-olds; and ‘older’: six‐eight-year-olds) and an adult comparison group. The evidence indicates a strong correspondence of adult and child responses to structural complexity, both in terms of global fluency measures and in terms of more detailed indicators of planning load. In addition, we report some specific contrasts in the patterning for children and adults that suggest disparities in processing resources and/or in local planning strategies. Children’s utterances are typically used to evaluate their knowledge of words and structures. Their utterances are less often studied for what they reveal about language processing per se or to address the e!ects of interactions between a performance system and the linguistic competence that

Elements of communicative content that are not expressed by constituents of the sentence uttered, what we will call "unexpressed elements of content" (UECs), played an important role in the history and development of generative grammar. In the 80's and 90's, mostly inspired by the work of Grice, UECs and mechanisms for recovering them not contemplated by linguistic theory of the time, began to surface under a variety of labels. We will collectively refer to these phenomena as 'impliciture' (extending Bach: 1994). Impliciture phenomena raise some interesting questions, only some of which is the field in a position to say anything about at present. Levinson (2000) has taken the useful step of distinguishing what we call "Q-phenomena", mostly related to linguistic scales, from "I-phenomena", mostly related to stereotypical information. Starting in the late 80's, experimental work began on the nature of these unspoken contents and their attendant mechanisms. In a recent study, Garrett and Harnish (2007) looked at I-phenomena, that have been proposed to depend on stereotypical background information. We asked whether these contents really are delivered by such mechanisms as "default heuristics", operating on general background knowledge, or whether they might be more tied to language via something like the "standardization" of a form for a certain use. Initial results for the materials tested seem to favor standardization as a mechanism for delivering this content. Completed research by Orjada (2007) and Rybarova (2007) extends the study of impliciture to additional examples and new populations. One assesses performance in RH damaged populations. The other contrasts performance for populations with high and low frontal lobe function.

Grice proposed to investigate ‘the total signification of the utterance’. One persistent criticism of Grice’s taxonomy of signification is that he missed an important category of information. This content, and/or the process of providing it, goes by a variety of labels: ‘generalized implicature’, ‘explicature’, ‘unarticulated constituents’, ‘default heuristics’, ‘impliciture’. In this study we first take a sample of such phenomena and, from the point of view of pure pragmatics, survey the central descriptions of the content expressed and the mechanisms that might deliver these contents. We then, from the point of view of experimental pragmatics, focus on two accounts: Levinson’s I-heuristic, and Bach’s standardization. We find experimental evidence for the existence of such implicitures, and for the use of language specific standardizations over language neutral background ­information.

Background: Most pragmatic abilities are relatively preserved in aphasia, while ability to express spatial relationships is compromised. Do aphasic speakers observe the pragmatic constraint that unexpected information should be expressed while inferrable information may be omitted?Aims: Asking aphasic speakers to describe unexpected spatial arrangements on objects should produce attempts at expressing spatial information that are likely to contain errors.Methods & Procedures: A total of 16 participants with various types of aphasia and 10 controls described pictures showing expected and unexpected arrangements of furniture.Outcomes & Results: As predicted, this pragmatic manipulation of stimulus materials produced large amounts of data for analysis: Participants tried to produce elaborated descriptions of unexpected arrangements, and these efforts contained many errors. Qualitative analysis showed that these errors were not perceptual or conceptual, yet involved encoding problems much broader than “diffic...

In the experiments outlined in this article, the authors investigate lexical access processes in language production. In their earlier work, T. Pechmann and D. Zerbst (2002) reported evidence for grammatical category constraints in a picture-word interference task. Although grammatical category information was not activated when subjects produced bare noun descriptions of simple objects, a robust effect arose when the target word had to be embedded in a syntactic frame. The current experiments demonstrate that compilation of a simple noun phrase (NP) yields word class effects in picture-word interference experiments in the same time frame as that generally observed for semantic processing. Most significantly, the effect emerges both in German and English with very similar profiles. On these grounds, it is implausible that the effect depends on syntactic gender activation, because such constraints are lacking in the English language version of the experiments.

Contains fulltext : 62733-OA.pdf (Publisher’s version ) (Open Access) We present a new technique for studying the activation of semantic and phonological codes in speech planning using event-related brain potentials (ERPs) that extend a well-established behavioral procedure from speech production research. It combines a delayed picture-naming task with a priming procedure. While participants prepared the production of a depicted object's name, they heard an auditory target word. If the prepared picture name and the target word were semantically or phonologically related, the ERP waveform to the target word tended less towards the negative when compared to an unrelated control. These effects were widely distributed. By contrast, if participants performed a nonlinguistic task on the depicted object (natural size judgment), the semantic effect was still obtained while the phonological effect disappeared. This suggests that the former effect indexes semantic activation involved in object processing while the latter effect indexes word-form activation specific to lexical processing. The data are discussed in the context of models of lexical access in speech production.

We report an experiment in which we test the relationship between gender and number in subject-predicate agreement. We also test the link between two different number-agreement relations—subject-verb and subject-predicative adjective. Participants saw first an unmarked adjective and then a sentence fragment consisting of a complex subject with a head noun and a modifier containing a second noun and were asked to make a whole sentence using the adjective with the proper gender and number markings. The gender of the subject head and the gender and number of the attractor noun were manipulated. Number errors in the verb and number and gender errors in the predicative adjective were measured. The results suggest gender agreement is computed independently of number agreement. In contrast, subject-verb number agreement and subject-predicative adjective number agreement are a unitary process. The implications for psycholinguistic and linguistic theories of gender and number are discussed.

This paper examines the role of semantic factors in the production of subject-verb number agreement. As an ostensibly grammatical process, number agreement provides an interesting case for examining the flow and interaction of semantic and syntactic information through the language-production system. Using a sentence-completion task, agreement errors can be elicited from subjects by presenting them with sentence fragments containing a complex noun-phrase, in which the nonhead noun is plural (e.g., The key to the cabinets ... WERE missing.). Previous research has demonstrated that the probability of making an error can be affected by varying the properties of the nouns in the complex noun phrase. By investigating which variables do and do not affect error rates, constraints on the flow of information through the production system can be inferred. In three experiments, we investigated the possible effects of three different semantic manipulations of the nouns in the complex NP: animacy, semantic overlap, and plausibility of modification by the sentence predicate. We found that both animacy and semantic relatedness had reliable effects on error rates, indicating that the mechanism involved in implementing agreement cannot be blind to semantic information. However, the plausibility with which each noun could serve as the subject of the sentence predicate had no effect on error rates. Taken together, these results suggest that while semantic information is visible to the agreement mechanism, there are still constraints on when this information can affect the process. Specifically, it may be the case that only information contained within the complex NP is considered for the purposes of implementing agreement.

To use a word in a sentence requires speakers to access information such as the meaning of the word, its use in a sentence (i.e., its syntax), and its sound form. In this study we examine whether when speakers cannot retrieve the sound form of a word, they are nevertheless able to indicate its syntactic properties. More specifically, we investigated whether English speakers in a tip of the tongue state, and a brain-damaged speaker in an anomic state, could correctly guess whether target words are either “count nouns” (e.g.,an opinion) or “mass nouns” (e.g.,some knowledge) when they could not say them. Results showed that speakers can correctly guess syntactic features such as count and mass, extending previous results concerning grammatical gender in Italian. We discuss the implications for models of language production in which lexical retrieval includes two steps. The first step involves retrieval of a word's abstract representation, specifying meaning and syntax. The second step involves the retrieval of the sound pattern of the word. Additional evidence from slips of the tongue in Italian, Spanish, and German showing syntactic constraints on phonological errors is also presented.

Deep dyslexia is an acquired reading disorder. Investigations of this disorder have focused primarily on the error pattern for single words read aloud. An essential characteristic of deep dyslexia is a profound inability to read closed class elements (‘function words’). In Garrett's language production model, closed and open class words (‘content words’) are retrieved by distinct mechanisms: Open class elements are retrieved independently of their phrasal environment, but closed class words are integral parts of phrase structure. This study's expectation was that if syntactic environments supporting closed class are available to deep dyslexic readers, their ability to read closed class words might improve. Two English speaking patients and one Italian speaking patient were tested using a text and list format manipulation. All three were significantly better at reading closed class words presented in text format than list format. Open class words that are generally read poorly (e.g. verbs) did not...

To correctly produce words, speakers must have access to three broad classes of information lexical semantics, syntax, and sound structure The relevant information must be organized in ways that permit rapid and accurate retrieval of specific lexical targets Current models of language production do this by a two-stage process The first stage incorporates lexical meanings and syntax, and the second, sound structure We used studies of the tip-of-the-tongue phenomenon (i e, the condition in which a speaker cannot produce a well-known word) to evaluate this organization and in so doing, we provide the first clear experimental evidence for a lexical stage that includes syntax and is distinct from both sound structure and the conceptual correlates of syntactic features.

We report basic findings on articulation rate effects derived from experimental research on children's sentence planning. The experiments were designed to evaluate production processes with attention to syntactic variables while controlling for lexical and phonological variables. Articulatory rate measures showed (1) rates differed for children (3–8.11) and adults, with adult rates significantly faster than child rates; (2) the effects of fluent and dysfluent host utterances on fluent substrings differed for adults and children; (3) rate patterns for relative clause and conjoined clause utterances differed, both from each other and across age group.

This article claims that two things will bulk large in the historical scene. One is the importation of brain and biology into cognitive science in general and psycholinguistics in particular. The other is more subtle, but equally significant: the embedding of language in more general theories of mind. There is a reawakening of interest in the evolutionary underpinnings of language and other characteristically human features of cognition; an interest obviously confounded with the increased emphasis on the neuroscience of language and cognition. One might roughly describe the changes in practices and attitudes toward psycholinguistic problems in terms of three periods: 1955–1970: shaking the shackles of behaviorism and establishing the field; 1970–1990: exploring the fundamental problems and finding new directions; 1990s–: building cognitive science and investing in new methods. The article also looks at the state of current research and some probable future directions in several areas, including lexical processing, speech perception, developmental psycholinguistics, parsing, connectionism, cross-linguistic emphasis, language production, bilingualism, and the neuropathologies and biology of language.

Purpose The Scales of Cognitive and Communicative Ability for Neurorehabilitation (SCCAN; L. Milman & A. Holland, 2007) was developed in the hospital setting to address changes in assessment practice. The SCCAN was designed to provide an overview of impairment and activity limitations across 8 cognitive scales (Speech Comprehension, Oral Expression, Reading, Writing, Orientation, Attention, Memory, and Problem Solving). The scales were developed using item response theory so that tailored testing could be implemented to reduce test administration time. This research investigated the validity and reliability of the SCCAN. Method A battery of neuropsychological tests was administered to 40 neurologically healthy control participants and 51 participants diagnosed with left-hemisphere pathology, right-hemisphere pathology, or probable Alzheimer’s disease. Analyses were performed to assess test sensitivity and specificity, construct validity, administration time, and reliability. Results The test accurately classified 95% of the control participants and 98% of the participants diagnosed with neurological disorders. Results indicate that the test also differentiated the performance profiles of the 3 clinical populations. In addition, test scores correlated significantly with external measures of the same cognitive areas. Mean administration time was 34 min. Test–retest stability ( r = .96, p < .001) and internal consistency ( r = .99, p < .001) coefficients were both significant, indicating that tailored testing procedures generated reliable test scores.

This paper presents the Featural and Unitary Semantic Space (FUSS) hypothesis of the meanings of object and action words. The hypothesis, implemented in a statistical model, is based on the following assumptions: First, it is assumed that the meanings of words are grounded in conceptual featural representations, some of which are organized according to modality. Second, it is assumed that conceptual featural representations are bound into lexico-semantic representations that provide an interface between conceptual knowledge and other linguistic information (syntax and phonology). Finally, the FUSS model employs the same principles and tools for objects and actions, modeling both domains in a single semantic space. We assess the plausibility of the model by showing that it can capture generalizations presented in the literature, in particular those related to category-related deficits, and show that it can predict semantic effects in behavioral experiments for object and action words better than other models such as Latent Semantic Analysis (Landauer & Dumais, 1997) and similarity metrics derived from Wordnet (Miller & Fellbaum, 1991).

This article is a revision of the previous edition article by M. Garrett, volume 22, pp. 14864–14870, © 2001, Elsevier Ltd.

Publisher Summary This chapter reviews the remarks on the architecture of language processing systems. The chapter highlights some of the recurrent conflicts among experimental outcomes that rely on behavioral indices of parsing. These controversies can be understood in terms of two additional data classes: a recent series of eye-movement studies that supports strong interactive claims on re parsing and an equally recent series of event related potential (ERP) studies that supports modular systems. The chapter analyzes the argument that the various conflicts between empirical findings require a resolution in terms of a filtering model, and sketches the outline of such a proposal. The essential feature of the resolution relies on using the language production system as the source of the apparent conceptual and discourse level constraints on parsing. Any general evaluation of the architecture of language processing systems must address not only the character of language comprehension, but also that of language generation. Language production models must account for the real-time integration of utterance form—in the service of the specific force—of the meaning that a speaker wishes to convey on a given occasion of utterance.

This paper reports studies of subject-verb agreement errors with speakers of Spanish and English; we used a sentence completion task, first introduced by Bock and Miller (1991). In a series of four experiments, we assessed the role of semantic information carried by the sentential subject in the induction of subject-verb agreement errors. For Spanish speakers, a sentence preamble such as la etiqueta sobre las botellas (the label on the bottles), which is usually interpreted to denote several labels, induced more agreement errors than preambles that normally denote a single entity. This finding replicates previous research with Italian (Vigliocco et al., 1995). English speakers, on the other hand, were not sensitive to this semantic dimension, as was found earlier by Bock and Miller (1991). This cross-linguistic difference is discussed in the framework of a modified version of the computational model of grammatical encoding proposed by Kempen and Hoenkamp (1987). In this version of the model agreement is computed through a unification operation instead of feature-copying, allowing for an independent retrieval of agreement features from the conceptual representation for the subject and the verb. We propose that languages differ in the extent to which the selection of the verb is controlled by features on the subject and features from the conceptual representation.

Foundations of Language Brain, Meaning, Grammar, Evolution. Ray Jackendoff. Oxford University Press, New York, 2002. 497 pp. $40, £25. ISBN 0-19-827012-7. Building on developments in linguistics over the past half century, Jackendoff offers an interpretation of "parallel architecture" to explain how the brain stores and processes language.

This is the second part of the JPR special issue "Processing of grammatical gender" documenting the Processing of Grammatical Gender workshop held in Leipzig, Germany, in July 1998. The special issue is organized into two sections. The first section, presented in Part I of JPR, has been devoted to papers on language comprehension. This part presents mainly work on language production. For the editorial of this special issue, please refer to Part I in the previous issue of JPR (Volume 28, Number 5). Part II is led into by a review article on the representation and processing of grammatical gender in language production (Schriefers & Jescheniak). The first empirical paper by Marx investigates the levels of processing that give rise to the gender-identity effect observed in German speech errors. Vigliocco and Zilli employed induced speech errors in Italian normal and aphasic speakers to investigate whether nonsyntactic—both conceptual and morphophonological—information is used in encoding the syntactic structure of a sentence and whether the integration of syntactic and nonsyntactic information can be differentially impaired in Broca's aphasics. Turennout et al. used motor-related brain potentials to investigate the time course of grammatical and phonological processes during speech production in Dutch. Their results provide evidence that grammatical processes precede phonological processes. Bentrovato et al. report interacting effects of semantic sentence context and grammatical gender on picture naming. Results are interpreted in support of interactive activation models of language processing. Akhutina et al. present four experiments establishing gender priming in Russian using an auditory word repetition naming paradigm. Hagoort & Brown present event-related brain potential evidence for the syntactic nature of grammatical gender processing during sentence comprehension. Two research notes close the special issue. In the first note, Jescheniak reports two experiments on Introduction to Part II of the Special Issue, "Processing of Grammatical Gender"

Research indicates that auditory and visual information is integrated during the perception of speech. Conflicting auditory and visual stimuli can result in an illusory experience known as the McGurk effect (e.g., auditory /bav/ dubbed onto a face saying /gav/ results in a perception of ‘‘dav’’). This study used a priming paradigm to investigate whether a phonemic representation for the auditory portion of a McGurk stimulus is active after the illusory phoneme is experienced. Subjects were given (nonword) prime‐target conditions, including: (1) McGurk (e.g., Prime auditory /bav/ + visual /gav/ = ‘‘dav;’’ Target auditory /bav/); (2) Incongruent (e.g., Prime auditory‐visual /mav/, Target auditory /bav/); (3) Identity (e.g., Prime auditory‐visual /yav/, Target auditory /yav/). Results show that mean reaction times to repeat targets were fastest in the identity condition. Response times for the McGurk and incongruent conditions were indistinguishable from one another and significantly slower than the identity condition. This finding suggests that once the auditory and visual information is combined and a phonemic representation is made, the actual auditory signal is no longer available to affect processing of the target. [This work is based on ideas developed by the late Kerry P. Green and supported by NSF.]

The fifteen original contributions in Language and Space bring together the major lines of research and the most important theoretical viewpoints in the areas of psychology, linguistics, anthropology, and neuroscience, providing a much needed synthesis across these diverse domains. The study of the relationship between natural language and spatial cognition has the potential to yield answers to vexing questions about the nature of the mind, language, and culture. The fifteen original contributions in Language and Space bring together the major lines of research and the most important theoretical viewpoints in the areas of psychology, linguistics, anthropology, and neuroscience, providing a much needed synthesis across these diverse domains. Each chapter gives a clear up-to-date account of a particular research program. Overall, they address such questions as: how does the brain represent space, how many kinds of spatial representations are there, how do we learn to talk about space and what role does culture play in these matters, should experimental tests of the relations between space and language be restricted to closed-class linguistic elements or must the role of open-class elements be considered as well? Throughout authors speak to each other's arguments, laying bare key areas of agreement and disagreement. Contributors Manfred Bierwisch, Paul Bloom, Melissa Bowerman, Karen Emmorey, Merrill Garrett, Ray Jackendoff, Philip Johnson-Laird, Barbara Landau, Willem Levelt, Stephen Levinson, Gordon Logan, Jean Mandler, Lynn Nadel, John O'Keefe, Mary Peterson, Daniel Sadler, Tim Shallice, Len Talmy, Barbara Tversky

Six Broca's and six Wernicke's aphasics were tested for their processing of prepositions in different linguistic tasks which placed increasing emphasis on the use of syntactic knowledge. Subjects performed a variety of tasks involving the production and perception of sentences containing prepositions. Agrammatics' production performance was very poor. Although their comprehension performance was superior to production when asked to complete a sentence selecting the target item out of four prepositions their comprehension performance was no longer superior to production in tasks where correct performance strongly depends upon the knowledge of sentence form. These results support the view of a general incapacity of agrammatics to use knowledge of sentence structure in production and in comprehension.

The perceptual complexity of lexically ambiguous and unambiguous sentences was compared in three experiments. In Experiment 1, the report of ambiguous words from rapidly presented ambiguous sentences was worse than the report of corresponding unambiguous words from unambiguous sentences. Results of Experiment 2 showed that the effect was not reduced by the presence of prior biasing context within the sentence. Experiment 3 repeated the finding with a sentence meaning classification task. It was concluded that both meanings of a lexically ambiguous sentence must be computed, even when prior context makes one meaning more plausible than the other.

Experimentation with 14–18-week-old infants indicates that they are capable of grouping together syllables of English depending on whether the syllables share a consonant. These results indicate that infants may have access to the mechanisms that underlie certain perceptual constancies in adult speech perception.

The effect of the lexical complexity of verbs on the processing of sentences was evaluated in two experiments. Verb complexity was indexed by the number of types of grammatical structure a verb permits (e.g., a verb may be transitive or intransitive and may permit various types of complement structures). Ss’ performances in paraphrasing sentences and in solving anagrams containing complex verbs were significantly poorer than their performances with the same sentences and anagrams containing less complex verbs.

The perceptual complexity of three lists of self-embedded sentences was evaluated in terms of the accuracy and time required for their paraphrase. The lists differed by the presence of relative pronouns in one list, their absence in a second and by the addition of adjectives to the third. It was predicted that the presence of the relative pronouns would effect the only significant change in performance. In both auditory and visual presentations of the sentence lists, the presence of the relative pronouns proved to be facilitating, while the presence of the adjectives produced no significant changes.

Judgments of the location of short bursts of noise in sentences were used to reveal perceptual segmentation of sentences. It was assumed that segmentation would correspond to major constituent boundaries. In order to control for correlated variables of pitch and intonation, identical acoustic material was provided with alternate constituent structures. It was found that differences in response to identical strings were predicted by the points of variation in constituent structure.

A lexical decision paradigm was used to examine syntactic influence on word recognition in sentences. Initial fragments of sentences were presented visually (CRT display) one word at a time (at reading speeds), from left to right. The string terminated with the appearance of a lexical decision target. The grammatical structure of the incomplete sentence affected lexical decision reaction time (RT). In Experiment 1, modal verb contexts followed by main verb targets and preposition contexts followed by noun targets produced lower RTs than did the opposite pairings (i.e., modal/noun and preposition/verb). In Experiment 2, transitive verb contexts followed by noun targets and subject noun phrase contexts followed by verb targets yielded lower RTs than did the opposite pairings. Similar contrasts for adjective targets did not yield comparable effects in Experiment 2, but did when the adjective was the head of a predictable phrase (Experiment 4). In Experiment 3, noun targets yielded lower RTs than did verb targets after contexts of a transitive verb followed by a prepositional phrase. An account of these effects is offered in terms of parsing constraints on phrasal categories.

The distinction between closed and open class words--of interest in the first instance because of claims about the support of structural analysis during comprehension--has its reflection in word recognition. For normal speakers, performances over these types give evidence of the operation of separate recognition devices, while for agrammatic speakers, performance indicates no such separation. A study of recognition accuracy with tachistoscopic presentation lateralized to the visual hemifields suggests some parallel between left and right hemisphere processing and the contrast of normal and agrammatic speakers: with direct input restricted to the left hemisphere, the vocabulary types show different levels of accuracy; no such differences are evident with presentations to the right.

Questions about the order of processing events during speech perception are among the most interesting and challenging in the psychology of language. Relating segmentation strategies to computational effort appears to be a possible approach to a difficult area. If we knew where processing takes place and where it is held in abeyance, we would have some basis for inferring what kinds of information are needed by the recognition system for the projection of grammatical structures. Knowing what kinds of structures are, in fact, inferred by the hearer at given points in the sentence would help us to refine our guesses about the function of specific kinds of information (e.g., acoustic, phonetic, lexical, etc.).